Novo 2.0
A 0-to-1 web app for building U.S. work visa petitions.
Founded in 2018, Legalpad was an immigration startup whose mission was connecting global talent with opportunity. We used a combination of technology and legal expertise to dramatically improve the U.S. immigration process for immigrants and their employers. I joined post-seed round in December 2018 as the 8th overall and first design hire.
After successfully improving our customer intake experience, I was tasked with improving Novo, our internal visa construction and management platform.
In 2018, traditional U.S. law firms were taking 4 to 6 months to build a competitive work visa application, or “petition”. The typical process went something like this:
The product team’s mission was to cut that process down to 2 months and eliminate or significantly reduce RFEs by improving Novo and the overall petition creation experience.
In July 2019, I traveled to Legalpad’s main office in Seattle and spent 3 days with the Operations team. Operators were primarily responsible for building petitions and comprised the primary Novo user base.
Interviews & Observations
I conducted interviews with each Operator and spent a day shadowing Kaitlin, our lead operator and Novo expert, as she worked through building a petition.
Journey Mapping Workshop
To better understand the holistic petition process and the people involved, I conducted a journey mapping workshop with the entire team (SMEs, Sales, and Operators). As expected, it was an ambiguous process with lots of moving parts.
Benchmarking the current system
Along with the interviews & observations, I conducted a survey to better understand the current pain points, and get a sense of Novo’s current usability. The survey yielded some interesting (and alarming!) results:
• Building petitions is a frustrating process, with “Document Generation” as the biggest Operator pain point.
• Time on task estimates varied wildly, indicating a lack of standardization and varying degrees of domain expertise.
• With a SUS score of 42.5, Novo was really hard to use.
While the Operator survey generated a ton of insightful data points, I found this quote in particular to be pretty eye-opening:
“i hate filing out the g-28, i-129, i-907. Like i absolutely despise this process its so tedious and so much going back and forth between novo and adobe where i am editing the the forms.”
Synthesis
When I got back to Denver, I synthesized the data I’d collected into four key problems with the current Operator tools and processes:
• It’s hard to know what to work on next.
• Some critical features don’t work or don’t appear to work.
• Critical flows are confusing and/or overly cumbersome.
• Novo is missing critical flows.
I worked with the engineering team to get the bugs prioritized, and started thinking about how we might come up with a better system.
Card Sort & IA Modeling
To better understand the problems I’d uncovered and the Operator mental model, I ran remote card sorting exercises with each Operator. It seemed like Novo had most of the pieces it needed, but they weren’t organized in a way that made sense to its users — the Information Architecture was off.
To help me think through this, I diagrammed the existing IA. After a few iterations, I landed on a model that seemed to best represent the root problem — information was siloed by type. You could see the forest, the trees, and even the weeds, but you couldn’t zoom in or out without losing your place.
This was especially problematic for Operators — they wanted to look at data points in the context of the document they were currently working on; looking at the data by type was overwhelming.
With that in mind, I proposed a new model that emphasized easily transitioning between lenses and zoom levels without losing context.
Sketching a better system
Once I had a model to help me think through the IA and some concrete ideas on how to restructure it, the engineering team & I started sketching the high-level flows & interaction patterns we thought might address Novo’s problems.
Wires
I wired up our sketches, and then quickly got operator feedback on a few of the early design concepts.
Concept Prototype
Using a series of impromptu co-design sessions with Operators, engineers, and SMEs (basically whoever had a few minutes), I fleshed out a high-level concept of the entire system. Working in extremely low fidelity, often just using words to describe an idea or feature, kept these sessions quick and welcoming to non-designers.
Visual Design explorations
Once I had enough signal that the overall concept was on the right track, I started exploring high-fidelity visual and interaction design options. Because building visa petitions is such a cognitively-heavy process, my explorations focused on minimizing cognitive load and maximizing scannability by:
• Using color and size to establish a clear, glancable hierarchy
• Employing distinct affordances — being really intentional when using the primary action color
• Establishing and resuing clear, actionable patterns
Another major consideration was the front-end implementation. Knowing that I would be developing the front-of-the-front-end, I limited my explorations to layouts and styles I was confident I could build quickly.
💀 “Kill your darlings”
I was stoked to the max very happy with how the Document Cards were shaping up, and couldn’t wait to start building them. As soon as I started putting them into realistic layouts, however, I noticed problems.
• Long document titles became difficult to read
• Petition sections with lots of documents became hard to scan
• Icon-only buttons are hard to learn and rarely usable
R.I.P., Document Cards. Hello, Document Rows!
Usability Test
Once the overall design felt good enough, I planned, prepped, and facilitated a round of formal usability testing. In general, the design performed well with a few moderate usability issues.
I synthesized the test results and presented them to the product team. We decided to start building Novo 2.0. 🎉
Story Mapping & Planning
To align on the Novo 2.0 vision and scope, I facilitated a slimmed down Pivotal Labs-style Inception Workshop. Because the engineering team was so involved in the design process and generally pretty rad, we moved through this 2-day workshop in about 4 hours.
Delivery Prototype
With the MVP scoped, I quickly built a prototype for the team to reference as we started building.
Coded Style Guide
I built a front-end component style guide and a few sample layouts for the engineers to use as a reference and grab code from. This eliminated the need for any design reviews or design-focused UAT (yuck), and made me very popular with the engineering team. 😊
A few weeks later, we released the Novo 2.0 MVP. 🚀
Surprise & Delight
While we didn’t have any quantitative indicators of success (we descoped product analytics from our MVP release), the customer feedback was immediate and overwhelmingly positive.
Super Screenings
By improving the petition creation workflow, we had created a powerful new tool for our Sales team and improved the onboaring experience for immigrants.
Long-term success
The Novo 2.0 MVP laid the foundation for Legalpad’s vision of “an operating system for work visas,” and its viability directly contributed to its long-term financial success:
• A few months after Novo 2.0’s release, Legalpad raised a $10M Series A.
• In 2022, Legalpad was acquired by Deel, Inc..